AI Model Wiki

    Skip to main content
    6 min read
    Lesson 2 of 5

    Using Le Chat & La Plateforme

    From the course Mistral AI: European Open Models

    Mistral offers two main interfaces: Le Chat for consumers and La Plateforme for developers.

    Le Chat (chat.mistral.ai)

    Le Chat is Mistral's consumer chatbot, similar to ChatGPT:

    • Free tier — Access to Mistral's models with generous limits
    • Canvas — Collaborative document editing with AI assistance
    • Web search — Real-time information access
    • Image understanding — Upload and analyze images with Pixtral
    • Code execution — Run Python code directly in the chat
    • Document analysis — Upload PDFs and documents for analysis

    Key Le Chat Features

    • Agents — Create custom AI assistants with specific instructions and knowledge
    • Flux image generation — Generate images directly in the chat
    • Multilingual — Excellent performance in European languages
    • Citations — Web search results include source citations

    La Plateforme (console.mistral.ai)

    For developers, La Plateforme provides:

    • API access to all Mistral models
    • Fine-tuning capabilities
    • Usage analytics and monitoring
    • Team management and billing
    • Playground for testing prompts

    API Quick Start

    from mistralai import Mistral
    
    client = Mistral(api_key="your-api-key")
    
    response = client.chat.complete(
        model="mistral-large-latest",
        messages=[{"role": "user", "content": "Explain quantum computing"}]
    )
    print(response.choices[0].message.content)
    

    The API also supports the OpenAI-compatible format for easy migration.

    Common Pitfall: Using the OpenAI-Compatible Endpoint Without Checking Feature Parity

    Mistral's API accepts OpenAI-format requests at https://api.mistral.ai/v1/chat/completions, but not every OpenAI feature maps to Mistral. Notably: logprobs, n (multiple completions), and some response_format options may behave differently or be unsupported. If you switch from the OpenAI SDK to Mistral using the base URL override, run your test suite against Mistral explicitly before considering the migration complete.

    Best Practice: Use Le Chat Agents to Prototype Prompts Before Committing to API Code

    Before writing API code, prototype your prompt and system instructions in Le Chat's Agents interface. Le Chat Agents let you configure a persistent system prompt, upload knowledge documents, and test the model's responses interactively. Once the prompt behaves correctly in Le Chat, copy the exact system prompt into your API code — this typically saves 2–4 rounds of prompt iteration in production.

    Named Product Reference: Mistral's OpenAI-Compatible Base URL

    // Drop-in replacement for the OpenAI SDK
    import OpenAI from "openai";
    
    const mistral = new OpenAI({
      apiKey: process.env.MISTRAL_API_KEY!,
      baseURL: "https://api.mistral.ai/v1",
    });
    
    const response = await mistral.chat.completions.create({
      model: "mistral-small-latest",
      messages: [{ role: "user", content: "Explain the Eiffel Tower in French" }],
    });
    

    This pattern lets you switch between OpenAI and Mistral by changing only two lines, making A/B testing across providers trivial.

    Key Takeaways

    • •Le Chat is Mistral's free consumer chatbot with web search and image features
    • •La Plateforme provides developer API access, fine-tuning, and analytics
    • •Le Chat Agents let you create custom AI assistants with specific knowledge
    • •The API supports both native Mistral SDK and OpenAI-compatible format
    0 of 5 lessons marked complete

    Progress is kept in this browser only.